555 EFI again...

[Chris Morriss]

In message <35A42575.5D66 at huron.net>, Clare Snyder <snyder at huron.net>
writes
>Chris Morriss wrote:
>> 
>> In message <35A14466.3AD1CF47 at brokersys.com>, Walter Petermann
>> <corsaro at brokersys.com> writes
>> >Frederic Breitwieser wrote:
>> >>
>> >> >I'm finished with the circuit and am working on the power supply.
>> >>
>> >> I would love a copy :)
>> >>
>> >
>> >Here it is:
>> >
>> >http://brokersys.com/~corsaro/wiring_diagrams/555fe.htm
>> >
>> >let me know what you think.
>> >
>> > Walter
>> I'm writing this before looking at this website, so perhaps this design
>> is OK.  The point I want to make is that the core of a simple EFI is the
>> voltage to pulse-width stage.  If you use a 555 and simply alter the
>> threshold level you won't get a linear v to PW relationship.  A normal
>> 555 monostable uses an R-C network for the timing.  The voltage across
>> the capacitor rises in an exponential fashion, whereas we require that
>> the voltage rises in a linear ramp.  This can be acheived by using a
>> constant current source to charge the cap.  Voltage across the cap will
>> vary according to dv/dt = i/C.
>> 
>> With this technique the desired pulse width can be set by altering the
>> threshold voltage, or by keeping a fixed threshold and making the
>> current source into a voltage controlled one.  This can be done by the
>> conventional op-amp and transistor VCCS, (ground referenced) with a
>> current mirror used to provide the current from the V+ rail.
>> --
>> Chris Morriss
>How do we implement this - modification to the referenced circuit?

Ah well there's the second problem.  The easiest way to make a pulse
that gets longer with an increase in the control voltage would be to
replce the resistor feeding the capacitor with a constant current
source, (to give a linear ramp), and then to vary the voltage at which
the internal comparator switches (This would be the control voltage from
the MAP sensor circuitry). The 555 does bring this voltage out to a
reference pin, (usually decoupled with a cap).  The problem is that this
reference voltage is internally supplied in the 555 from a resistor
chain at a voltage of 2/3 of Vcc. (ie 3.33 V on a 5V supply).  It is
possible to overide this voltage by driving the pin from an external
voltage source, but the internal Vref for the trigger input is fed from
the same divider chain, and altering the voltage externally affects the
trigger operation.
I've no idea why the 555 designers didn't use two separate voltage
references for the threshold and the trigger.
Have a look at the schematic of the inside of a 555 on the .PDF file on
the NAT. Semi web site. (www.national.com,  do a search for 555).

There may be timers around where you can set the voltage at which the
device triggers, but I've not found a 555 variant that allows this.  I
ended up making a voltage to pulse-width stage by using a voltage
comparator (LM311 will do), a D-type flip-flop, and a N-channel FET to
discharge the capacitor.  (There was also a constant current source
sourcing current from the Vcc rail)

I think that I'll have to do a crude circuit diagram of a suitable stage
and email out a .BMP (It won't take many K bytes)

I can't manage this before next week though.

Hope most of you understand the hardware techy bits above.
-- 
Chris Morriss